Subgroup analysis

Subgroup analysis refers to repeating the analysis of a study within subgroups of subjects defined by a subgrouping variable (e.g. smoking status defining two subgroups: smokers and non-smokers).[1]

Aim

The aim of subgroup analysis is usually to assess whether the association of two variables differs depending on a third variable. For instance, investigators of a study testing the effect of an intervention (variable 1) on preventing heart attacks (variable 2) might be interested in whether the effect varies by smoking status (variable 3). Therefore, they could perform separate analyses for smokers and non-smokers, and then compare the results. If results differ, there is a subgroup effect.

Terminology

The terminology is inconsistent. Alternative names for subgrouping variables include effect modifiers, predictive factors, or moderators. Alternative names for subgroup effects include effect modification and interaction. Some authors define effect modification and interaction differently, and distinguish different types of interaction. A more unspecific term is heterogeneity of treatment effects.

gollark: Raffling off one of these because I was suddenly reminded of the existence of their parents: https://dragcave.net/lineage/dAv9i
gollark: Of course there was a raffle, someone started a raffle.
gollark: Wait, how do I actually know what the raffle is *for*?
gollark: !raffle enter
gollark: I meant Iridium and such. Starlink is too heavy and power hungry.

See also
  • Post-hoc analysis

References
  1. Lagakos SW (20 April 2006). "The Challenge of Subgroup Analyses — Reporting without Distorting". NEJM. 354 (16): 1667–9. doi:10.1056/NEJMp068070. PMID 16625007.


This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.